MadGraph5_aMC@NLO

Calculating cross-section in monojet search

Asked by Yu-Xiang Lin

Dear MadGraph team,

I am testing a model for dark matter monojet research, I generate p p > DM DM j, with new fermions running in the t-channel.
I only need the value of cross-section, but I noticed that when I generate for different number of events(nevents = 1,10,100,1000,10000), MadGraph gave me different value of cross section, for example: when nevents=1 it gave me about 1e-4, nevents=10000 it gave me about 1e3, it seems unreasonable. I also got different cross section if I use different random seed. Aother things is the "nb_event_after matching" are not match the nevents I ask MadGraph to generate( sometimes it matches though).

Then I try another effective model with same process with new fermions in t-channel, and found the same problem.

After doing some check, I found that if my new fermions mass above around 3TeV, the I got consistent cross section for different nvents. Or, if I generate p p > DM DM (without extra jet), it also got the consistent cross section

I have no clue about this situation, can you give me some advise?

Thank you.

Kind regards,
Sean

Question information

Language:
English Edit question
Status:
Answered
For:
MadGraph5_aMC@NLO Edit question
Assignee:
No assignee Edit question
Last query:
Last reply:
Revision history for this message
Olivier Mattelaer (olivier-mattelaer) said : 		#1

Hi,

The main problem seems that you do not know what is a matching/merging computation and how such computations works.

Let me stress the basic and refer you to our wiki for more details information.

1) You have double counting between p p > DM DM and p p > DM DM j. (due to PDF and as renormalization)
2) To avoid such double counting, we use MLM as default (you can also decide to use CKKW-L)
3) In MLM, we use alpha_s and PDF re-weighitng to modify the matrix-element to behaves like the parton-shower in soft area. However we do NOT multiply by the Sudakof form factor --which will remove the double counting (only CKKW method is doing it and no one is using that method)
4) So the Sudakof form factor is implemented in MLM has an efficiency cut within the parton-shower.

So this is why you can not only generate only one phase-space points to compute your cross-section since this is not enough to have a nice estimate of the efficiency of the accept/reject cut related to the Sudakof form-factor. (and therefore why your cross-section can change quite a lot when you do not ask that many events)

Cheers,

Olivier

Can you help with this problem?

Provide an answer of your own, or ask Yu-Xiang Lin for more information if necessary.

To post a message you must log in.